Bug 173901
| Summary: | kasan disables kmemleak | ||
|---|---|---|---|
| Product: | Other | Reporter: | CAI Qian (caiqian) |
| Component: | Other | Assignee: | other_other |
| Status: | RESOLVED CODE_FIX | ||
| Severity: | normal | CC: | catalin.marinas, dvyukov, luto |
| Priority: | P1 | ||
| Hardware: | All | ||
| OS: | Linux | ||
| Kernel Version: | 4.8-rc8 | Subsystem: | |
| Regression: | No | Bisected commit-id: | |
|
Description
CAI Qian
2016-09-29 19:25:54 UTC
This warning seems to be generated by depot_alloc_stack() and not related to kmemleak. Is there any message from kmemleak in the kernel log which says why it was disabled? In general kmemleak_disable() call whould be accompanied by a pr_warn() with more information. So I retested it this on today's mainline HEAD. Although kmemleak remains enabled after bootup, the kernel won't be able to survive any kernel compilation workload. [ 130.392553] kasan: CONFIG_KASAN_INLINE enabled [ 130.397517] kasan: GPF could be caused by NULL-ptr deref or user memory access [ 130.405581] general protection fault: 0000 [#1] PREEMPT SMP KASAN [ 130.412382] Modules linked in: intel_rapl sb_edac edac_core x86_pkg_temp_thermal intel_powerclamp coretemp kvm_intel kvm irqbypass crct10dif_pclmul crc32_pclmul ghash_clmulni_intel aesni_intel lrw gf128mul glue_helper ablk_helper cryptd iTCO_wdt iTCO_vendor_support pcspkr mxm_wmi i2c_i801 i2c_smbus sg mei_me mei lpc_ich shpchp ipmi_ssif ipmi_si ipmi_msghandler wmi acpi_power_meter acpi_pad nfsd auth_rpcgss nfs_acl lockd grace sunrpc ip_tables xfs libcrc32c sr_mod cdrom sd_mod mgag200 i2c_algo_bit drm_kms_helper syscopyarea sysfillrect sysimgblt fb_sys_fops crc32c_intel ttm ixgbe drm serio_raw ahci libahci mdio libata ptp i2c_core pps_core dca fjes dm_mirror dm_region_hash dm_log dm_mod [ 130.480996] CPU: 82 PID: 799 Comm: kmemleak Tainted: G B 4.8.0+ #3 [ 130.489156] Hardware name: Intel Corporation S2600WTT/S2600WTT, BIOS GRRFSDP1.86B.0271.R00.1510301446 10/30/2015 [ 130.500513] task: ffff880824298000 task.stack: ffff8808242a0000 [ 130.507121] RIP: 0010:[<ffffffff81677ee0>] [<ffffffff81677ee0>] scan_block+0xb0/0x460 [ 130.515976] RSP: 0018:ffff8808242a7d20 EFLAGS: 00010046 [ 130.521905] RAX: dffffc0000000000 RBX: ffff880824298000 RCX: 0000000000000001 [ 130.529868] RDX: 1ffff10104854fc8 RSI: 0000000000000000 RDI: ffff8808242a7e40 [ 130.537830] RBP: ffff8808242a7d78 R08: 0000000000000001 R09: 0000000000000001 [ 130.545793] R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000007ff9 [ 130.553756] R13: 1ffff10104853368 R14: ffff880824298000 R15: 0000000000000000 [ 130.561720] FS: 0000000000000000(0000) GS:ffff880e5ed80000(0000) knlGS:0000000000000000 [ 130.570749] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 130.577161] CR2: 00007f65eceae990 CR3: 0000000002e0a000 CR4: 00000000003406e0 [ 130.585124] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [ 130.593086] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 [ 130.601049] Stack: [ 130.603292] 0000000000000000 ffffffff82e090d8 ffff880824299b40 ffff880e4994c9c0 [ 130.611589] 0000000000000292 0000000000000000 ffff8807fb08c200 dffffc0000000000 [ 130.619886] ffff8807fb08c200 0000000001080000 ffffea0042000000 ffff8808242a7de8 [ 130.628179] Call Trace: [ 130.630910] [<ffffffff81679907>] kmemleak_scan+0xb37/0xf50 [ 130.637119] [<ffffffff81678e08>] ? kmemleak_scan+0x38/0xf50 [ 130.643435] [<ffffffff81679d20>] ? kmemleak_scan+0xf50/0xf50 [ 130.649849] [<ffffffff81679d9e>] kmemleak_scan_thread+0x7e/0xd0 [ 130.656554] [<ffffffff811f6e02>] kthread+0x222/0x2e0 [ 130.662182] [<ffffffff811f6be0>] ? kthread_park+0x80/0x80 [ 130.668304] [<ffffffff811f6be0>] ? kthread_park+0x80/0x80 [ 130.674424] [<ffffffff811f6be0>] ? kthread_park+0x80/0x80 [ 130.680548] [<ffffffff826f93ba>] ret_from_fork+0x2a/0x40 [ 130.686571] Code: df 41 0f b6 44 05 00 84 c0 74 08 3c 03 0f 8e 84 03 00 00 4c 89 fe 83 ab 40 1b 00 00 01 48 b8 00 00 00 00 00 fc ff df 48 c1 ee 03 <80> 3c 06 00 0f 85 30 03 00 00 49 8d be 40 1b 00 00 48 be 00 00 [ 130.708363] RIP [<ffffffff81677ee0>] scan_block+0xb0/0x460 [ 130.714591] RSP <ffff8808242a7d20> [ 130.718536] ---[ end trace 7f1b35189505276b ]--- [ 130.723688] Kernel panic - not syncing: Fatal exception [ 131.794896] Shutting down cpus with NMI [ 131.799190] Kernel Offset: disabled [ 131.803108] ---[ end Kernel panic - not syncing: Fatal exception [ 131.809833] ------------[ cut here ]------------ [ 131.814992] WARNING: CPU: 82 PID: 799 at arch/x86/kernel/smp.c:125 native_smp_send_reschedule+0x89/0xa0 [ 131.825476] Modules linked in: intel_rapl sb_edac edac_core x86_pkg_temp_thermal intel_powerclamp coretemp kvm_intel kvm irqbypass crct10dif_pclmul crc32_pclmul ghash_clmulni_intel aesni_intel lrw gf128mul glue_helper ablk_helper cryptd iTCO_wdt iTCO_vendor_support pcspkr mxm_wmi i2c_i801 i2c_smbus sg mei_me mei lpc_ich shpchp ipmi_ssif ipmi_si ipmi_msghandler wmi acpi_power_meter acpi_pad nfsd auth_rpcgss nfs_acl lockd grace sunrpc ip_tables xfs libcrc32c sr_mod cdrom sd_mod mgag200 i2c_algo_bit drm_kms_helper syscopyarea sysfillrect sysimgblt fb_sys_fops crc32c_intel ttm ixgbe drm serio_raw ahci libahci mdio libata ptp i2c_core pps_core dca fjes dm_mirror dm_region_hash dm_log dm_mod [ 131.894064] CPU: 82 PID: 799 Comm: kmemleak Tainted: G B D 4.8.0+ #3 [ 131.902222] Hardware name: Intel Corporation S2600WTT/S2600WTT, BIOS GRRFSDP1.86B.0271.R00.1510301446 10/30/2015 [ 131.913578] ffff880e5ed87cb0 ffffffff81a6a6e1 0000000000000000 0000000000000000 [ 131.921874] ffff880e5ed87cf8 ffffffff8118e3f2 0000000000000046 ffffffff0000007d [ 131.930171] 0000000000000003 0000000000000003 0000000000022fc0 0000000000022fc0 [ 131.938466] Call Trace: [ 131.941193] <IRQ> [<ffffffff81a6a6e1>] dump_stack+0x85/0xc4 [ 131.947630] [<ffffffff8118e3f2>] __warn+0x172/0x1b0 [ 131.953170] [<ffffffff8118e61d>] warn_slowpath_null+0x1d/0x20 [ 131.959681] [<ffffffff810efa79>] native_smp_send_reschedule+0x89/0xa0 [ 131.966967] [<ffffffff8126b1c3>] trigger_load_balance+0x563/0xa10 [ 131.973865] [<ffffffff8126ada7>] ? trigger_load_balance+0x147/0xa10 [ 131.980956] [<ffffffff812275c4>] scheduler_tick+0x1b4/0x300 [ 131.987276] [<ffffffff8134e9d0>] ? tick_sched_do_timer+0xf0/0xf0 [ 131.994082] [<ffffffff8131f7a7>] update_process_times+0x47/0x60 [ 132.000785] [<ffffffff8134d139>] tick_sched_handle.isra.16+0x49/0xe0 [ 132.007972] [<ffffffff8134ea39>] tick_sched_timer+0x69/0xe0 [ 132.014288] [<ffffffff81321c96>] __hrtimer_run_queues+0x2f6/0xbf0 [ 132.021186] [<ffffffff81323d7e>] ? hrtimer_interrupt+0x13e/0x480 [ 132.027987] [<ffffffff813219a0>] ? retrigger_next_event+0x150/0x150 [ 132.035076] [<ffffffff81323d7e>] ? hrtimer_interrupt+0x13e/0x480 [ 132.041877] [<ffffffff81323de2>] hrtimer_interrupt+0x1a2/0x480 [ 132.048485] [<ffffffff810f7c53>] local_apic_timer_interrupt+0x73/0xf0 [ 132.055771] [<ffffffff826fbf3b>] smp_apic_timer_interrupt+0x7b/0xa0 [ 132.062861] [<ffffffff826fafd6>] apic_timer_interrupt+0x96/0xa0 [ 132.069562] <EOI> [<ffffffff81503a6b>] ? panic+0x2d1/0x311 [ 132.075895] [<ffffffff81503a64>] ? panic+0x2ca/0x311 [ 132.081532] [<ffffffff8150379a>] ? percpu_up_read_preempt_enable.constprop.33+0xb9/0xb9 [ 132.090566] [<ffffffff812d184d>] ? kmsg_dump+0x22d/0x310 [ 132.096592] [<ffffffff8109a5f0>] oops_end+0xc0/0xd0 [ 132.102131] [<ffffffff8109aabb>] die+0x4b/0x70 [ 132.107187] [<ffffffff8109499f>] do_general_protection+0x20f/0x3b0 [ 132.114183] [<ffffffff826fa668>] general_protection+0x28/0x30 [ 132.120694] [<ffffffff81677ee0>] ? scan_block+0xb0/0x460 [ 132.126718] [<ffffffff81679907>] kmemleak_scan+0xb37/0xf50 [ 132.132937] [<ffffffff81678e08>] ? kmemleak_scan+0x38/0xf50 [ 132.139253] [<ffffffff81679d20>] ? kmemleak_scan+0xf50/0xf50 [ 132.145666] [<ffffffff81679d9e>] kmemleak_scan_thread+0x7e/0xd0 [ 132.152369] [<ffffffff811f6e02>] kthread+0x222/0x2e0 [ 132.158005] [<ffffffff811f6be0>] ? kthread_park+0x80/0x80 [ 132.164127] [<ffffffff811f6be0>] ? kthread_park+0x80/0x80 [ 132.170248] [<ffffffff811f6be0>] ? kthread_park+0x80/0x80 [ 132.176369] [<ffffffff826f93ba>] ret_from_fork+0x2a/0x40 [ 132.182392] ---[ end trace 7f1b35189505276c ]--- The latest kernel dump seems to be unrelated to the original one. I can indeed trigger it with the latest kernel but I'm not sure it depends on KASan. It seems that task_stack_page() returns NULL occasionally (maybe caused by the recent vmap stack changes on x86; I'll add some debug printks to kmemleak and re-test). Commit 68f24b08ee89 ("sched/core: Free the stack early if CONFIG_THREAD_INFO_IN_TASK") causes the task->stack to be set to NULL. I'll have to check if this happens while read_lock(&tasklist_lock) but from the commit log, it seems that it might be possible.
I didn't do anything to intentionally prevent it. The right way to fix it should be to use try_get_task_stack(). I fixed it in this patch: http://lkml.kernel.org/r/1476266223-14325-1-git-send-email-catalin.marinas@arm.com If Cai is no longer seeing any issues, we could close this bug. Yes, everything works fine now. CAI, please close this bug as it's fixed now. This still shows up as STATUS:NEW bug for KASAN. |